I'm sorry.... but, if you read a few current science books, you will find out that everything I've described is not only possible, but a lot of it isn't even especially new.
We seem to have agreed that there is "stuff" in many recordings above 20 kHz.
And that it is NOT "undifferentiated random noise" at all.
Rather, it is a combination of the interaction with real sounds and the acoustics of a real room, almost certainly mixed in with some actual random noises and some distortion.
(If you even glance at the file BigShot provided, it's pretty obvious that what's there isn't at all random; it clearly follows some sort of patterns - and patterns are another word for information.)
For starters.....
All I need is one distinct wavefront which I can analyze (so I pick out a loud drumbeat).
By measuring the arrival times between the initial wavefront, and the first three or four echoes, I will now know the dimensions of my room.
Actually, I'll have to do this same analysis with several sounds, and correlate the results.
In 1950 this would have required a really expensive computer; nowadays the processor in my cheap cell phone is quite capable of doing it in a few seconds.
And they use it for everything from autonomous cars, to ultrasound imaging, to ground penetrating RADAR...
(And, yes, they figured out how to read a RADAR pulse that's way below the noise floor a long time ago....)
So, if I wanted to use it, I would simply continue on from where we already are.
I would use a high pass filter to separate the ultrasonic content from the "audible music".
Then I would boost it a bit to make it easy to work with.
Then I would fed it into a DSP engine running appropriate software to analyze it.
However, you are consistently exaggerating the difficulties and complications involved in the basic process.
How do we differentiate the useful information from the random noise?
That's easy.
We all know what white noise looks like..... on an oscilloscope or a spectrum display....
I took those high-passed files BigShot thoughtfully provided...
All I had to do was to turn up the volume.
It was quite obvious that a lot of it wasn't actually random... and simple to pick out the interesting parts.
I could easily identify one loud click, pick out a few return echoes that follow it, and get my room dimensions from there.
Of course, a DSP doing spectral analysis will extract more useful information than I can by simply looking at it, but you get the idea.
Now, is it really USEFUL to figure out that the snare drum was recorded in a plywood isolation booth ten feet square, while the violins were recorded in a concert hall?
Interesting question there.
Obviously it IS information that we didn't have before.
Perhaps I'll list it on my new expanded album folder.
Or, perhaps, I'll describe it in my new book: "The provenance of historical drum track recordings".
Or, just maybe, I'll write new software that cleans up recordings, by removing conflicting room acoustic cues from the drum tracks.
(I could pick out and remove the echoes of the drum from that crummy little plywood booth... and replace them with calculated reverb that matches the reverb on the violins.)
Our brains use cues like that as part of how they visualize the sound stage.
Perhaps the recording sounds a bit jumbled because the drum sounds include cues about room acoustics that conflict with the cues I'm hearing from the violins.
So, perhaps, once I know that, by detecting and "cleaning up" the drum hits, I can make the whole recording sound better.
(If I was really enthusiastic I would include an algorithm to add fake room acoustics to the drum track that matches the violin track.)
How "unreasonable" or "unlikely" is all this to be useful - especially to a home user?
Well, now, I really don't know.
You might be surprised to know how much calculation is involved in synthesizing a center channel, or height channels, from a stereo recording.
Izotope's latest restoration utility has the ability to remove distortion, clipping, and harmonics it can recognize.
(Can it recognize them more accurately if we include those ultrasonic harmonics?)
It also has the ability to not just add, but to
REMOVE, reverb....
Apparently it is able to pick out the specific sounds of reverb, differentiate them from primary sounds, and reduce their level - after the fact.
There are also several new pieces of software that allow you to "match the acoustics of one track to another" - presumably by analyzing, and then collectively replicating, the "echoes" and "room noise".
(And, no, I have no idea which of the current batch benefit from having extra ultrasonic information... but, of course, that could change tomorrow anyway.)
I would simply prefer to preserve as much of "what might be useful information" - in case it DOES turn out to be useful.
And, no, any
competently designed piece of audio gear will not suffer from excessive IM distortion because the bandwidth of the incoming signal is too wide.
(It is standard design practice, in a proper design, to specifically limit the bandwidth of the input stage to block any frequencies that would cause problems at later stages.)
And, yes, give me $1 million, and I will show you a flying pig (remember the Doritos commercial)....
Of course, it would be more humane to simply buy my pet pig a seat next to me the next time I fly to Paris....
And, finally...... obviously that 90 kHz carrier tone
DOES exist - or the Plangent process wouldn't work.
In fact, anyone who actually knows how the technology of recording magnetic tape works already
KNEW that it existed.
It's been there all along... but a lot of people simply
weren't paying attention for a long time.
1. Not happy with only misrepresenting the facts, now you're misrepresenting the responses to your misrepresentations. That's not an uncommon tactic by audiophiles who are newbies to this (or other science/fact based) forums but you're not a newbie here. So what on earth made you think that ridiculous tactic would work here and specifically on me??
1a. You're joking? All you've done is misrepresented the facts and then invented hypothetical theories/suggestions based on those misrepresented facts, the word "hypocrite" seems wholly inadequate!!
2. I have NOT disputed there might be something above 20kHz, that would be absurd as I've recorded "something" above 20kHz thousands of times, over a period of 20 years or so. And, I've clearly acknowledged that tape bias (at about 90kHz) exists and that ultrasonic content exists from record clicks. So, you are MISREPRESENTING both what I believe and what I've actually stated! Containing "something" above 20kHz and containing acoustic information above 20kHz are two entirely different things though, two different things which you are fallaciously correlating!
3. No, the usual assertion is that's it's worse than useless. Being inaudible makes it typically useless but potentially causing IMD, which is audible, makes it worse than useless!
4. What "reality" is that? Is that an existing audiophile "reality" or one you've just made-up yourself?
4a. Just to be clear about your "reality", you're saying that: We've got some random ultrasonic noise being produced and instead of being absorbed by the walls/boundaries it's reflected (let's say because the walls are made of sheets of diamond). These (hypothetical) reflections must, according to the laws of physics, be substantially lower than our original random noise but you go way further and propose that we then calculate the RT60, the amount of time it takes for those random noise reflections to decay by a further 60dB. Let's use some figures: Let's say we've got ultrasonic random noise at -40dBFS and a diamond walled studio producing random noise reflections at say -60dBFS. How do you propose we differentiate the -60dB random noise that's underneath the -40dB random noise? Then, we calculate the RT60 of the reflections, the amount of time it takes for our (hypothetical) -60dB reflections to decay to -120dB. How do you propose to differentiate -120dB random noise that's underneath -40dB random noise? So, a diamond walled studio and an impossible differentiation process is your "reality" is it? In other words, flying pigs exist but they're invisible, stealth flying pigs that we can't see or detect. If that's not all absurd enough, you go even further! ...
4b. Not satisfied with the impossible task of just differentiating (theoretical) random noise that's way underneath other random noise, you now propose that we can actually analyse that undifferentiatable random noise and "probably calculate" the acoustic properties of our diamond walled studio. In the real world (rather than your "reality") we can't even achieve this feat with reflections that do actually exist and are above the noise. In fact, the latest software is only partially successful at determining what is a reflection from what is direct sound, let alone analyse and get any detailed acoustic information from it. So now we've got invisible, stealth flying pigs that don't exist and can't be seen or detected but with magic we can determine all kinds of complex details such as; how fast they're flying, their altitude, where they've flown from and where they're heading. But wait, there's even more! The assertion that this non-existant, non-detactable, non-analysable, non-information is actually useful! If we ignore the fact that it doesn't exist and we can't detect or analyse it and pretend that we can determine the details of where it was recorded, how does that help in anyway? In virtually all commercial audio recordings going back many decades, we actively try to change/disguise the room and acoustics of where it was recorded. Let's take an obvious example to illustrate the point. Let's say we've got a film scene in a big hall in a castle. Most probably it was actually filmed on a film set made of plywood or in a small studio with a green screen, some of the dialogue was probably recorded in a voice-over booth/ADR suite, all the other sounds were manufactured or recorded in various other places, a Foley studio and other locations. How, even if it were possible, is acquiring this information useful? We want it to sound like it's in a castle hall, not a plywood set and a combination of various different studios! The same applies to music production, since the mid-late 1960's onwards, pretty much none of it is supposed to sound like the actual combination of locations in which it was recorded. The only potential exception would be recording a symphony orchestra or some purely acoustic performance in a single venue (such as a concert hall), using only a coherent mic array, say only an ambisonics setup. So, you're talking about roughly 0% of all commercial audio recordings in the last 50 odd years!
5. How will I reduce the data we can get from it if there is no data we can get from it? How can you start with nothing and then reduce it?
6. They are neither remotely practical nor examples! But apart from that your sentence is spot on!
6a. How is that an example or in any way analogous to the situation? It would be analogous if vinyl record clicks contained no information >20kHz and then the obvious question would be: How well would your ultrasonic click detector work then?! And as for being practical, that's so ridiculous, it's funny. Virtually all your hypothesising on this topic is about what some future technology might be able to achieve, you think maybe the future is some sort of high definition, surround sound vinyl LP?
6b. Likewise, if that 90kHz bias tone didn't actually exist, how effective would the Plangent process be then? And again, how is this (non-analogous) example even remotely practical, how many consumers have even heard of it, let alone actually use it to correct wow and flutter on studio master tapes?
1. That's an utterly FALSE misrepresentation! Even if we assume it is actual instrument harmonics above 20kHz, that would still provide
NO INFORMATION about the acoustics of the recording venue. What *might* provide that information would be the REFLECTIONS of those harmonics from the recording venue's boundaries (walls, floor, ceiling). So my question (still) is, where's your evidence that there are any reflections above 20kHz? Outside your fantasy diamond walled studio, real concert halls and recording venues typically employ wood and other acoustic materials specifically to absorb mid/high freqs to produce a "warm" sound and the absorption coefficients of these materials INCREASES WITH FREQUENCY, additionally, so does the absorption by air. In the real world then, not only do we have low, very low or no harmonics above 20kHz to start with but what does exist is then NOT reflected off the walls, because the walls are doing the exact opposite and absorbing rather than reflecting them! So, thanks for reposting bigshot's >20kHz screenshot, now all you have to do is TAKE THE FIRST STEP and point out where the reflections/acoustic information is!!!!!
2. What "multiple arrival times"? The "multiple arrival times" of the >20kHz reflections which don't exist? Enough of this flying pig BS, show us some evidence that >20kHz reflections exist in music/audio!
3. AGAIN, stop just making up BS! Outside of your invented "reality", the actual analysis "possible these days" couldn't even work out that it is a cave!!
4. More utter BS! In the real world and with commercial audio, ABSOLUTELY NONE of that can be "done with audible content"! How then do you get "better resolution on the measurements" in a frequency range where that acoustic information doesn't even exist??
You're right, as far as actual hearing music content goes, stuff above 15 khz is always going to be extremely marginal, IMHO. Actually this is super-easy to hear (4 khz) (it's where your hearing sensitivity is greatest):
